New study finds bigger datasets might not always be better for AI models

From ChatGPT to DALL-E, deep learning artificial intelligence (AI) algorithms are being applied to an ever-growing range of fields. A new study from University of Toronto Engineering researchers, published in Nature Communications, suggests that one of the fundamental assumptions of deep learning models—that they require enormous amounts of training data—may not be as solid as once thought.

ai models algorithms artificial artificial intelligence assumptions bigger chatgpt communications dall dall-e datasets deep learning engineering fields intelligence machine learning & ai nature researchers study toronto university university of toronto